The coefficient of interest is β2, the coefficient of the triple interaction term.The estimator in Eq.(4) is called the difference-in-difference-in-differences (DDD) estimate, and the set of coefficients β2 therefore captures the differential effects of the banking reform in different regions/cities.This triple difference approach also serves as a robustness test of our baseline model in (3).In the next section, we report the results from the standard DD setting as our baseline and discuss the results of Eq.(4) as heterogeneous effects across regions/cities.